adaptive wiener filtering approach for speech enhancement - copy
TRANSCRIPT
ADAPTIVE WIENER FILTERING APPROACH FOR SPEECH
ENHANCEMENT
INTERNAL GUIDE PRESENTED BY Ms.K.RENU SAHITYA KIRANAsst.Professor 1220409126
Objective
•To design and implement the speech enhancement techniques wiener filtering and adaptive wiener filtering .
Introduction
•Speech enhancement is one of the most Important topics in speech signal processing.•Speech enhancement aims to improve speech quality
by using various techniques.•Speech enhancement techniques• WIENER FILTER • ADAPTIVE WIENER FILTER
Applications of speech enhancement
• Improving quality and intelligibility (hearing aids, cockpit comm., video conferencing ...)
• Source coding (mobile phone, video conferencing, IP phone ...)
• Pre-processor for other speech processing applications (speech recognition, speaker verification ...)
WIENER FILTER
• The basic principle of the Wiener filter is to obtain a clean signal from that corrupted speech signal.
•A Wiener can be an IIR or FIR. •Filter coefficients are calculated to minimise the
average squared distance between the filter output and a desired or target signal.
The filter input out put relation is given by
Is the wiener filter coefficient vectorThe Wiener filter error signal, e(m), is defined as the difference between the desired
(or target) signal, x(m), and the filter output
e(m) = x(m) - = x(m) The wiener filter coefficients are obtained by minimising an average squared error
function Έ[e²(m)], with respect to the filter coefficient vector, w, where Έ is average the mean square estimation is given by
Έ[e²(m)] =
Wiener filter for De-Noising Speech
The output of a Wiener filter is given by
The Wiener filter coefficient vector
W= Ryy⁻¹ryx
for uncorrelated speech and noise, the wiener filter equation can be
written as w=( Rxx + Rnn )⁻¹rxx
Noisy signal is
Y(f)=X(f)+N(f)
•The frequency Wiener filter obtained as
•Dividing num & den of above eq with noise power
spectra Pnn(f) and substituting the variable SNR(f)=Pxx(f)/Pnn(f) yields
Applications of Wiener filter
• It reduces broadband additive noise.•Radar system identification• echo cancellation•Signal restoration•Signal restoration• In communication channel equalization .
ADAPTIVE WIENER FILTER
•The application of the Wiener filter in an adaptive manner is speech enhancement.•The adaptive Wiener filter is implemented in time
domain rather than in frequency domain•An adaptive filter is a digital filter that has self-
adjusting characteristics.• It is capable of adjusting its filter coefficients
automatically to adapt the input signal via an adaptive algorithm.
LMS Algorithm• computationally simpler version of the gradient search method is
the LMS filter.• where the error signal e(m) is the difference between the
adaptive filter output and the target(desired) signal x(m), given by
• the LMS adaptation equation: w(m+1) = w(m) + µ [y(m)e(m)] • The main advantage of the LMS algorithm is its simplicity both in
terms of the memory requirement and the computational complexity.
• The convergence rate of the filter coefficients depends on the choice of the adaptation step size µ .
RLS Algorithm• The RLS filter has a relatively fast rate of convergence to the optimal filter
coefficients.• Input signals: y(m) and x(m)• Error signal equation: e(m) = x(m) - (m-1)y(m) • filter coefficients adaption
• filter gain vector update
• Here 𝜆 is the adaption or forgetting factor and as in the range 0 > 𝜆 > 1
PSNR results in dB for speech enhancement approaches
Noisy Speech Wiener Filter Adaptive Wiener Filter
(LMS)
9.7855 12.5555 17.3821
10.4425 16.1351 21.7890
PSNR results in dB for speech enhancement approaches
Noisy Speech Wiener Filter Adaptive Wiener Filter (RLS)
10.1218 12.6027 17.5297
10.4333 16.1275 22.0635
CONCLUSION
• An adaptive Wiener filter approach for speech enhancement approach depends on the adaptation of the filter transfer function from sample to sample based on the speech signal statistics(mean and variance). This results indicates that the proposed approach provides the best SNR improvement compare to traditional Wiener filter approach in frequency domain. The results also indicate that the proposed approach can avoid the drawbacks of Wiener filter in frequency domain
REFERENCESY. Hu and P. Loizou: A subspace approach for enhancing speech
corrupted by colored noise, in Proc. International Conference on Acoustics, Speech and Signal Processing, vol. I,Orlando, FL, U.S.A., pp. 573-576, May (2002).
A. Rezayee and S. Gazor: An adaptive KLT approach for speech enhancement, IEEE Trans. Speech Audio Processing, vol. 9, pp. 87-95Feb. (2001).
Advance digital signal processing and noise reduction by Professor Saeed V. Vaseghi .
Digital signal processing fundamentals and applications by Li Tan .